GPS Transportation Mode Inference Using Acceleration and Transition Probabilities
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods fail to accurately infer a user's mode of transportation from raw GPS data, especially when multiple modes are used in a single trip and under varying conditions like traffic or weather, due to limitations in velocity-based rules and the need for additional sensor data.
Innovation Solution
A method that processes GPS data by segmenting trips into walk and non-walk segments based on velocity and acceleration, using an inference model to predict the most probable mode of transportation for each segment, and applying post-processing to adjust probabilities based on transition probabilities between modes.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of operation
If velocity-based rules are used to determine transportation mode, then the determination process is simple, but accuracy deteriorates when traffic conditions or weather cause driving velocity to be as slow as walking
Solution Approach 1:
The patent transforms the single velocity parameter into multiple parameters including velocity, acceleration, segment duration, and transition probabilities. By analyzing combinations of these parameters rather than relying solely on velocity thresholds, the system can distinguish between walking and slow driving even when velocities are similar, thereby improving determination accuracy while maintaining operational simplicity.
Solution Approach 2:
The system employs post-processing that uses transition probabilities between transportation modes to adjust initial predictions. By incorporating feedback from adjacent segments and considering the likelihood of mode transitions, the system refines its determinations and corrects errors that would occur with simple velocity-based rules alone.
2Measurement precision
If additional sensor data (cellular phone devices, Wi-Fi, RFID) are collected to improve accuracy, then transportation mode determination accuracy improves, but device complexity and data correlation requirements increase
Solution Approach 1:
The patent enables the GPS device to determine transportation mode using only its own internal sensors (accelerometer, velocity calculations from GPS positions). By making the GPS device self-sufficient and eliminating dependencies on external sensors, map data, or additional devices, the system achieves accurate determination without increasing overall device complexity or requiring complex data correlation between multiple sources.
Solution Approach 2:
The system uses the GPS device's existing components (position tracking, accelerometer) for multiple purposes: determining transportation mode, calculating velocity, and identifying segments. This multi-functional use of existing hardware avoids the need for additional specialized sensors while maintaining determination accuracy.
3Measurement precision
If manual tagging of GPS tracks with transportation information is implemented, then data accuracy improves, but user effort and time requirements increase
Solution Approach 1:
The system automatically determines and tags transportation modes without requiring user intervention. By using the device's own sensors and algorithms to infer transportation mode from GPS and accelerometer data, the system provides accurate transportation information automatically, eliminating the need for users to manually tag their trips while preserving data accuracy.
4Productivity
If simple velocity-based mechanisms are used, then processing speed is fast, but the ability to handle multiple transportation modes in a single trip deteriorates
Solution Approach 1:
The patent divides trips into segments based on changes in motion patterns detected by the accelerometer and velocity variations. By segmenting the trip data and analyzing each segment independently with transition probability models, the system can identify multiple transportation modes within a single trip while maintaining efficient processing. This segmentation approach allows the system to handle complex multi-mode trips without sacrificing processing speed.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
Described is a technology by which raw GPS data is processed into segments of a trip, with a predicted mode of transportation (e.g., walking, car, bus, bicycling) determined for each segment. The determined transportation modes may be used to tag the GPS data with transportation mode information, and/or dynamically used. Segments are first characterized as walk segments or non-walk segments based on velocity and/or acceleration. Features corresponding to each of those walk segments or non-walk segments are extracted, and analyzed with an inference model to determine probabilities for the possible modes of transportation for each segment. Post-processing may be used to modify the probabilities based on transitioning considerations with respect to the transportation mode of an adjacent segment. The most probable transportation mode for each segment is selected.